基于颜色特征的面料花型图案检索算法  

Fabric pattern retrieval algorithm based on color feature

作　　者：邱笑迎 Qiu Xiaoying(School of Computer Science and Technology(School of Artificial Intelligence),Zhejiang Sci-Tech University,Hangzhou,Zhejiang 310018,China)

机构地区：[1]浙江理工大学计算机科学与技术学院(人工智能学院),浙江杭州310018

出　　处：《计算机时代》2023年第9期92-95,共4页Computer Era

基　　金：浙江省科学技术厅“领雁”研发攻关计划项目(2022C01220)。

摘　　要：为使纺织企业更高效地管理海量的面料花型图案,提出一种基于图案主色及其聚集区域和分块颜色矩相结合的花型图案检索算法。首先利用改进的聚类算法提取花型图案的主颜色,同时根据主颜色提取主色的聚集区域特征;然后提取分块图案的颜色矩来表示局部颜色特征;最后对三种特征进行加权融合。在自建的面料花型图案库中进行实验,平均查准率达到85.8%,表明此方法能够有效地检索出色彩风格相似的图案,优于现有方法。In order to help textile enterprises manage massive fabric patterns more efficiently,a pattern retrieval algorithm based on the combination of pattern main color,its aggregation region and block color moments is proposed.Firstly,the improved clustering algorithm is used to extract the main color of the fabric pattern,and the aggregation region features of the main color are extracted based on the main color.Then,the color moments of block patterns are extracted to represent the local color features.Finally,the three features are weighted and fused.The experiments in the self-built fabric pattern database show that the average precision reaches 85.8%,which indicates that this method can effectively retrieve patterns with similar color styles and is superior to the existing methods.

关 键 词：花型图案 颜色特征 主颜色 颜色矩 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]